About

The name is meant to understate it. The reports are written the way the old annuals were, and underneath each one sits the kind of analysis that used to be the property of a front office: expected goals, shot locations, five-on-five on-ice results, league percentiles, a matchup model that widens rather than sharpens for the playoffs, and a workflow for making a call on a game. An armchair is a perfectly good place from which to know what you are talking about.

This site follows the format of the old pro hockey handbooks: a bio block, the numbers, and then a plain-spoken account of how a player actually plays — the finesse game, the physical game, the mental game, and the intangibles.

A new edition every season. Nothing is overwritten: each season's read stays up, so you can see what was said about a player three years ago and judge how it aged.

Why it is built like a risk system

This site is the second thing built out of one idea. The first was a credit surveillance platform for commercial real estate — the kind of system that watches a book of loans and reports what moved, what you are exposed to, and what it does not know. Hockey wanted the same machinery, because hockey is having the same argument: the analyst with the model against the man who has watched the asset for twenty years and can tell you something the model has no column for.

Credit surveillance has an answer to that argument, and the answer is not to pick a side. Six of its habits are the reason this site looks the way it does.

Awareness, not judgment

A surveillance system does not tell you to sell. It tells you that something moved, by how much, and against what baseline, and it leaves the decision where it belongs. The reports work the same way: the numbers narrow the question and the written read answers it, and neither is allowed to impersonate the other.

The divergence is the product

When the model and the scout disagree, the worst available move is to delete one of them. That disagreement is the most informative event in a credit file — the borrower whose ratios are clean and whose building is visibly wrong, or the reverse — and it is the most informative thing about a hockey player too. So where the prose and the numbers pull apart, this site shows both pulling rather than quietly resolving it.

Say what the model has never seen

Eighty-two games calibrate a rating. A seven-game series is a stress event the rating was never fitted on — different opponent quality, different deployment, different everything. So the playoff model widens rather than sharpens. The honest response to a regime your data does not cover is a wider band, not a more confident number, and a system that gets more certain exactly where it knows least is the failure mode this whole approach exists to avoid.

Beat the naive rule, or publish that you did not

Every model here is scored against something you could do without it. For the playoffs that bar is simply picking the home club in every series — a rule you can apply in your head. A model that cannot clear it is worth nothing, and the scoreboard goes up either way. As it happens, one of the two ratings on this site currently loses to that rule, and you can read exactly where on the record.

Know when the field entered the tape

In credit you learn quickly that a column existing is not the same as a column meaning something, and that the date a field entered the data is a fact you have to hold. The league's feeds return a value for all 1,759 team-seasons on record, but shots were not counted until 1959-60 and goals scored four-on-four were not separated out until 2009-10 — and several of those fields are written as zero rather than left empty before they begin, so they read as data when they are not. Every one of those dates is computed from the feeds themselves rather than remembered, and the check runs before anything is divided by anything.

Exposure rather than prediction

The most useful line in a surveillance report is rarely a forecast. It is a statement of what you are holding, so every club here is described by where its scoring actually comes from: the share of it produced by men thirty-three and older, the share produced by men under twenty-six, the share produced by men the club drafted itself, and how much of last season's five-on-five ice time is still on the roster at all. Those are exposures rather than projections, and they are worth more precisely because they are true today instead of probably true later.

None of which makes hockey a loan book. Nobody is underwriting anything here and the stakes are a pool entry. But the discipline transfers exactly: state the baseline, show the divergence, widen where you are ignorant, and never let the model's confidence outrun what the data can support.

How the reports are written

The numbers are pulled from public league data. Everything underneath them is an opinion — formed from watching the player and from how he was actually deployed, and revised as he changes.

Report length follows how much is genuinely known. A first-line centre gets the full treatment; a fourth-liner with forty games gets a short, honest paragraph about his role rather than invented detail padded out to match. Where the evidence is thin, the report says so.

Sources and rights

Statistics and biographical details are facts taken from the National Hockey League's public data feeds. Deployment — who played with whom — is derived from published shift charts by measuring shared ice time.

Photographs come from Wikimedia Commons under licences that permit reuse with credit; each one is credited on the player’s page, and all of them are listed on the photo credits page. Where no freely-licensed photograph exists, a player is shown by his sweater number.

This site is not affiliated with, endorsed by, or sponsored by the National Hockey League or any of its clubs. Team and league names are the trademarks of their respective owners and are used here only to identify the clubs a player plays for. No league photography, logos or other league artwork is reproduced.

All written evaluation, and the design of this site, are original work and belong to its author.

If you own material you believe has been used improperly here, get in touch and it will be removed.